MATH 436 Notes: Homomorphisms. Jonathan Pakianathan September 23, 2003 1 Homomorphisms Definition 1.1. Given monoids M1 and M2, we say that f : M1 → M2 is a homomorphism if (A) f(ab)= f(a)f(b) for all a, b ∈ M1 (B) f(e1)= e2 where ei is the identity element in Mi, i =1, 2. Basically a homomorphism of monoids is a function between them that preserves all the basic algebraic structure of a monoid: the binary operation and the identity. The function f : N → N given by f(n) = 0 for all n ∈ N is not a homomorphism of the monoid (N, ·) to itself even though condition (A) is satisfied. This is because 1 is the identity for multiplication and f(1) = 0 so condition (B) is not satisfied. Thus functions satisfying (A), do not automatically satisfy (B) when dealing with monoids. The situation with groups is somewhat different! Definition 1.2. Given groups G1,G2 a function f : G1 → G2 is called a homomorphism if f(ab)= f(a)f(b) for all a, b ∈ G1. One might question this definition as it is not clear that a homomorphism actually preserves all the algebraic structure of a group: It is not apriori obvious that a homomorphism preserves identity elements or that it takes inverses to inverses. The next proposition shows that luckily this is not actually a problem: Proposition 1.3. If f : G1 → G2 is a homomorphism between groups then: (1) f(e1)= e2 where ei is the identity element of Gi, i =1, 2. −1 −1 (2) f(x )=(f(x)) for all x ∈ G1. Thus f takes inverses to inverses. 1 Proof. (1): We compute f(e1) = f(e1e1) = f(e1)f(e1). Multiplying both −1 sides of this equation by f(e1) on the left we see that e2 = f(e1) as desired. −1 −1 (2): We compute using (1) that e2 = f(e1) = f(xx ) = f(x)f(x ). Thus −1 e2 = f(x)f(x ) for all x ∈ G1. Multiplying both sides of this equation by −1 −1 −1 f(x) on the left we find f(x) = f(x ) for all x ∈ G1 as desired. There are various special names for homomorphisms with certain prop- erties which we define next: Definition 1.4. Let f : M1 → M2 be a homomorphism of monoids (or groups). Then: →֒ a) If f is injective we call it a monomorphism. We typically write f : M1) M2 in this case. (b) If f is surjective we call it an epimorphism. We typically write f : M1 / /M2 in this case. =∼ (c) If f is bijective we call it an isomorphism. We typically write f : M1 → M2 in this case. Note in this case f sets up a one-to-one correspondence between the points of M1 and the points of M2 in such a way that the operation in M1 then corresponds to the operation in M2. We say M1 and M2 are isomorphic, denoted M1 ∼= M2 and we regard them as the same monoid (group) in different disguises. (d) A homomorphism f : M → M of a monoid (group) M back to itself is called an endomorphism of M. (e) A bijective endomorphism of M is called an automorphism of M. We consider some examples: Example 1.5. Let det : Matn(R) → R be the determinant function. Since det(AB)= det(A)det(B) and det(I)=1 in general, we see that det : Matn(R) → (R, ·) is a homomorphism of monoids where Matn(R) is a monoid under matrix multiplication. The determinant function restricts to also give a homomorphism of groups × × det : GLn(R) → (R , ·) where (R , ·) denotes the group of nonzero real numbers under multiplication. It is easy to check that det is an epimorphism which is not a monomorphism when n> 1. Let tr : Matn(R) → R be the trace function. Since tr(A + B)= tr(A)+ tr(B) we see that the trace gives a homomorphism tr : (Matn(R), +) → 2 (R, +) where Matn(R) and R are considered groups under addition. Again, it is easy to check that tr is an epimorphism which is not a monomorphism when n> 1. Example 1.6. Let A be an alphabet and let W (A) be the monoid of finite words on this alphabet under the operation of concatenation. Define L : W (A) → N by L(w) =length of the word w. Thus if A is the standard roman alphabet, L(cat)=3. It is clear L(ˆe)=0 where eˆ is the empty word and it is also easy to check that L(w1 ⋆ w2) = L(w1) + L(w2) and so L : W (A) → (N, +) is a homomorphism of monoids. It is called the length homomorphism. If |A| ≥ 2 then it is easy to check that the length homomorphism is an epimorphism but not a monomorphism. For example if a, b ∈ A are distinct then L(ab)= L(ba)=2 but ab =6 ba ∈ W (A). Example 1.7. Let I : Σn → Σn+1 be defined as follows: For σ ∈ Σn we set: σ(j) when 1 ≤ j ≤ n I(σ)(j)= (n +1 when j = n +1. Basically I takes a permutation σ of {1, 2,...,n} and extends it to a per- mutation of {1, 2,...,n,n +1} by just sending n +1 to itself. It is a simple exercise left to the reader to show that I is a monomorphism. Thus it in- duces an isomorphism between Σn and its image Im(I) ⊆ Σn+1. We often view Im(I) as a copy of Σn sitting within Σn+1. The following is a basic fact about homomorphisms: Theorem 1.8. (a) Let f : M1 → M2, g : M2 → M3 be homomorphisms of monoids (groups) then g ◦ f : M1 → M3 is also a homomorphism. (b) The identity map 1M : M → M is a homomorphism. (c) If S is a monoid (group) then End(S)= {f : S → S|f an endomorphism } is a submonoid of M(S), the monoid of functions with domain and codomain S. End(S) is called the monoid of endomorphisms of S. (d) If S is a monoid (group) then Aut(S)= {f : S → S|f an automorphism } is the group of units of End(S). It is a subgroup of Σ(S), the group of per- mutations on S. Aut(S) is called the group of automorphisms of S. Proof. (a): We compute (g ◦ f)(ab)= g(f(ab)) = g(f(a)f(b)) = g(f(a))g(f(b))=(g ◦ f)(a)(g ◦ f)(b) 3 and also (g ◦ f)(e1)= g(f(e1)) = g(e2)= e3 and so g ◦ f is a homomorphism of monoids (groups). (b): 1M (ab)= ab =1M (a)1M (b) and 1M (e)= e so 1M is a homomorphism of monoids (groups). (c): By (a) and (b), it follows that End(S) is a monoid under composition of functions and that it is a submonoid of M(S)= {f : S → S}. (d): Also follows directly as in (c). Before we can calculate a basic example of End(G) or Aut(G) we record a basic fact: Proposition 1.9. If G =<S> for some subset S of G and f1, f2 : G → H are two homomorphisms to another group such that they agree on S, i.e., f1(s) = f2(s) for all s ∈ S, then f1 = f2. In other words a homomorphism is uniquely determined by what is does on a generating set. ±1 ±1 ±1 Proof. Take g ∈ G. Since S generates G we have g = s1 s2 ...sk for some elements sj ∈ S. We then compute: ±1 ±1 f1(g)= f1(s1 ...sk ) ±1 ±1 = f1(s1) ...f1(sk) as f1 is a homomorphism ±1 ±1 = f2(s1) ...f2(sk) as f1 and f2 agree on S = f2(g) as f2 is a homomorphism. Thus f1(g)= f2(g) for all g ∈ G and so f1 = f2 and we are done. Proposition 1.10. Let (Z, +) be the group of integers under addition. Then End((Z, +)) = {fm|m ∈ Z} where fm : Z → Z is multiplication by m, given by fm(n)= mn for all n ∈ Z. Thus Aut((Z, +)) = {f−1, f1}. Furthermore the monoid End((Z, +)) is isomorphic to (Z, ·), the monoid of integers under multiplication. Aut((Z, +)) is thus isomorphic to the 2- element group of units of this monoid, i.e., {−1, 1}. ′ ′ Proof. Each fm is an endomorphism of (Z, +) as fm(n + n )= m(n + n )= ′ ′ mn + mn = fm(n)+ fm(n ). Suppose g is some other endomorphism of Z, then g(1) = m for some m ∈ Z. However fm(1) = m also and (Z, +) =< 1 > so by Proposition 1.9 we have that g = fm since they agree on the generating set. Thus this shows End((Z, +)) = {fm|m ∈ Z}. It is easy to check that the only fm : Z → Z that are bijections are f−1 and f1 so Aut((Z, +)) = {f1, f−1}. 4 Finally define θ :(Z, ·) → End((Z, +)) by θ(m)= fm. Since fm ◦fk = fmk and f1 =1Z it follows that θ is a homomorphism of monoids. It is trivial to check that it is a bijection and so induces an isomorphism between (Z, ·) and End((Z, +)). This completes the proof. The following is an important concept for homomorphisms: Definition 1.11. If f : G → H is a homomorphism of groups (or monoids) and e′ is the identity element of H then we define the kernel of f as ker(f)= {g ∈ G|f(g)= e′}. The kernel can be used to detect injectivity of homomorphisms as long as we are dealing with groups: Theorem 1.12 (Kernels detect injectivity).